Based on Kaggle's Freesound General-Purpose Audio Tagging Challenge from 2018: https://www.kaggle.com/c/freesound-audio-tagging/
This is a model composition that infers the source of a sound (trumpet, meow, etc.) given audio input, using the Keras library in Python. TagAudio.ipynb is the golden notebook.